www.nature.com/npjbcancer

ARTICLE OPEN Assessment of HMGA2 and PLAG1 rearrangements in breast adenomyoepitheliomas

Fresia Pareja1, Felipe C. Geyer1, David N. Brown1, Ana P. Martins Sebastião1, Rodrigo Gularte-Mérida1, Anqi Li1, Marcia Edelweiss1, Arnaud Da Cruz Paula1, Pier Selenica1, Hannah Y. Wen1, Achim A. Jungbluth1, Zsuzsanna Varga2, Juan Palazzo3, Brian P. Rubin4, Ian O. Ellis5, Edi Brogi1, Emad A. Rakha5, Britta Weigelt1 and Jorge S. Reis-Filho1

Breast adenomyoepitheliomas (AMEs) are rare epithelial-myoepithelial neoplasms that may occasionally produce myxochondroid matrix, akin to pleomorphic adenomas (PAs). Regardless of their anatomic location, PAs often harbor rearrangements involving HMGA2 or PLAG1. We have recently shown that the repertoire of somatic genetic alterations of AMEs varies according to their (ER) status; whilst the majority of ER-positive AMEs display mutually exclusive PIK3CA or AKT1 hotspot mutations, up to 60% of ER-negative AMEs harbor concurrent HRAS Q61 hotspot mutations and mutations affecting either PIK3CA or PIK3R1. Here, we hypothesized that a subset of AMEs lacking these somatic genetic alterations could be underpinned by oncogenic fusion , in particular those involving HMGA2 or PLAG1. Therefore, we subjected 13 AMEs to RNA-sequencing for fusion discovery (n = 5) and/or fluorescence in situ hybridization (FISH) analysis for HMGA2 and PLAG1 rearrangements (n = 13). RNA-sequencing revealed an HMGA2-WIF1 fusion in an ER-positive AME lacking HRAS, PIK3CA and AKT1 somatic mutations. This fusion gene, which has been previously described in salivary gland PAs, results in a chimeric transcript composed of exons 1–5ofHMGA2 and exons 3–10 of WIF1. No additional in-frame fusion genes or HMGA2 or PLAG1 rearrangements were identified in the remaining AMEs analyzed. Our results demonstrate that a subset of AMEs lacking mutations affecting HRAS and PI3K pathway-related genes may harbor HMGA2-WIF1 fusion genes, suggesting that a subset of breast AMEs may be genetically related to PAs or that a subset of AMEs may originate in the context of a PA. npj Breast Cancer (2019) 5:6 ; https://doi.org/10.1038/s41523-018-0101-7

INTRODUCTION and PAs, we sought to determine whether a subset of AMEs, Breast adenomyoepitheliomas (AMEs) are rare neoplasms with primarily those lacking mutations affecting known drivers (e.g., dual epithelial-myoepithelial differentiation,1 composed of gland- HRAS or PI3K pathway-related genes), would be genetically related like structures containing an inner layer of pink, eosinophilic to PAs, and would be underpinned by fusion genes, in particular epithelial cells and an abluminal layer of often clear, myoepithelial those involving HMGA2 and PLAG1. cells. AMEs can display a variety of histologic appearances, and be either estrogen receptor (ER)-positive or ER-negative.1,2 Although there is overlap in the histologic features of ER-positive and ER- RESULTS negative AMEs, we have recently shown that the repertoire of HMGA2-WIF1 fusion gene in an ER-positive AME genetic alterations of these tumors vary according to their ER Thirteen breast AMEs, whose whole-exome, targeted capture and/ 3 status. Whilst ER-negative AMEs harbor HRAS Q61 hotspot or Sanger sequencing and ER status were previously described in mutations co-occurring with mutations affecting PIK3CA or PIK3R1 Geyer et al,3 were included in this study (Table 1). Six cases were in up to 60% of cases, the majority of ER-positive AMEs were ER-negative and seven were ER-positive. Four ER-negative AMEs found to display seemingly mutually exclusive PIK3CA or AKT1 harbored concurrent PIK3CA and HRAS mutations (4/6), one activating hotspot mutations.3 harbored an HRAS Q61K mutation and concurrent likely patho- In the spectrum of histologic appearances of AMEs, myxochon- genic PIK3R1 mutations (1/6), and one was HRAS wild-type and 4 droid matrix has been occasionally described. This type of matrix harbored a PIK3CA mutation (1/6). Five ER-positive AMEs harbored bears histologic resemblance to the matrix of pleomorphic PIK3CA mutations (5/7), and all were wild-type for HRAS. None of adenomas (PAs),1 epithelial-myoepithelial neoplasms that may the cases harbored mutations affecting the AKT1 E17 hotspot arise in various anatomic locations, including the breast.5 PAs are locus (Fig. 1a and Supplementary Fig. 1). Notably, all HRAS and underpinned by recurrent gene rearrangements involving HMGA2 PIK3CA mutations were classical activating hotspot mutations, or PLAG1 in up to 65% of cases, regardless of their anatomic except for one PIK3CA mutation (Q546) which targeted a hotspot origin.6–8 Due to the overlapping histologic appearances of AMEs residue and was predicted to be likely pathogenic (Fig. 1a).

1Department of Pathology, Memorial Sloan Kettering Cancer Center, New York, NY, USA; 2Institute of Surgical Pathology, University Hospital Zurich, Zurich, Switzerland; 3Department of Pathology, Thomas Jefferson University Hospital, Philadelphia, PA, USA; 4Department of Pathology, Cleveland Clinic, Cleveland, OH, USA and 5Department of Pathology, University of Nottingham, Nottingham, UK Correspondence: Fresia Pareja ([email protected]) or Jorge S. Reis-Filho (reisfi[email protected]) These authors contributed equally: Fresia Pareja, Felipe C. Geyer Received: 12 September 2018 Accepted: 27 November 2018

Published in partnership with the Breast Cancer Research Foundation F. Pareja et al. 2

Table 1. Clinicopathological and selected genetic features of the 13 breast adenomyoepitheliomas included in this study

Case ID ER status PIK3CA status AKT1 PIK3R1 status HRAS Oncogenic fusion genes by RNA PLAG1/ HMGA2 rearrangement by status status sequencing FISH

AM1 Negative MUT WT WT MUT NT Negative AM4 Negative MUT WT WT MUT NT Negative AM8 Negative MUT WT WT MUT NT Negative AM11 Negative MUT WT NT MUT NT Negative AM5 Negative WT WT MUT MUT NT Negative AM7 Negative MUT WT WT WT None Negative AM2 Positive MUT WT WT WT None Negative AM3 Positive MUT WT WT WT None Negative AM12 Positive MUT WT NT WT NT Negative AM13 Positive MUT WT NT WT NT Negative AM17 Positive MUT WT NT WT NT Negative AM16 Positive WT WT NT WT HMGA2-WIF1 Positive (HMGA2) AM6 Positive WT WT WT WT None Negative ER estrogen receptor, MUT mutant, NT not tested, WT wild-type

To determine whether AMEs lacking HRAS Q61 hotspot likely pathogenic in-frame fusion gene was identified in the cases mutations would harbor fusion genes, we subjected five HRAS subjected to RNA-sequencing analysis (Supplementary Table 1).

1234567890():,; wild-type AMEs with available material to RNA-sequencing Given that HMGA2 and PLAG1 rearrangements have been analysis for an unbiased detection of expressed fusion genes described in other neoplasms with epithelial-myoepithelial (Supplementary Fig. 1). Using a validated pipeline for the de novo differentiation, in particular in PAs, we sought to define whether discovery of fusion genes,9 we identified an HMGA2-WIF1 fusion AMEs may harbor fusion genes known to underpin PAs. We gene in an ER-positive HRAS-/PIK3CA-/AKT1-wild-type AME (AM16) subjected the five AMEs analyzed by RNA-sequencing and all the = (Figs. 1a-c, Table 1 and Supplementary Table 1). The HMGA2-WIF1 other AMEs included in this study (n 8) to FISH using HMGA2 fusion gene identified in AM16 results in a chimeric transcript and PLAG1 dual-color break apart probes (Table 1 and Supple- fi encompassing all five exons and the initial segment of the 3′ UTR mentary Fig. 1). This analysis con rmed the presence of an HMGA2 of HMGA2 fused to exons 3–10 of WIF1, and is predicted to be rearrangement in both the epithelial and myoepithelial compo- nents of AM16 (Fig. 2b) and did not reveal any additional AMEs translated to a full length HMGA2 and an N-terminal harboring HMGA2 or PLAG1 rearrangements (Table 1). truncated WIF1 protein, with a truncated WIF domain. WIF1 encodes for a tumor suppressor that modulates Wnt signaling, a 10 role that requires an intact WIF domain. DISCUSSION The AME found to harbor the HMGA2-WIF1 fusion gene (AM16) 3 We have previously shown that approximately 60% ER-positive displayed the typical histologic features of AMEs, and constituted AMEs harbor PIK3CA or AKT1 mutations, whereas up to 60% of ER- a well-circumscribed lesion with pushing borders, surrounded by a negative AMEs are characterized by HRAS Q61 mutations fi thick brous capsule. This lesion displayed a nodular architecture concurrent with mutations affecting genes of the PI3K signaling and a mixed tubular and papillary growth pattern. No cellular pathway.3 Given the occasional histologic similarities between atypia, mitotic activity or necrosis was identified (Fig. 1c). Focal AMEs and PAs, and the fact that a subset of AMEs lack a known areas with conspicuous stroma with myxoid quality were observed driver genetic alteration in the form of somatic mutations (Fig. 1c). The myoepithelial component was highlighted by p63 on affecting protein coding genes, we posited that a subset of AMEs immunohistochemical analysis, and strong ER expression was may harbor oncogenic fusion genes previously described in other observed in the epithelial component (Fig. 1c). myoepithelial lesions including PAs. Our analyses resulted in the Given that in the salivary glands, epithelial-myoepithelial identification of an ER-positive HRAS-/PIK3CA-/AKT1-wild type AME carcinomas, the salivary gland counterpart of breast AME, can harboring an HMGA2-WIF1 fusion gene, which has been described occasionally originate in the context of a PA (i.e., the so-called in PAs and carcinomas ex-PA of the salivary gland.10,16 carcinoma ex-PA),11 we sought to define if AM16 would have The high-mobility group AT-hook 2 (HMGA2) gene encodes for a areas diagnostic of PA. An independent pathology review of all transcriptional regulator of genes involved in cell proliferation and 17 slides available from this AME by five pathologists failed to reveal cell death. HMGA2 overexpression plays a key role in oncogenic any areas that would be consistent with a diagnosis of PA. transformation through several mechanisms, such as the induc- tion of and AP1 activity, promotion of cyclin A expression, inactivation of -dependent apoptosis, and activation of the Breast AMEs lack recurrent HMGA2 or PLAG1 rearrangements 17,18 TGF-β signaling pathway. The Wnt signaling pathway is None of the additional AMEs subjected to RNA-sequencing regulated by secreted antagonists that bind to Wnt , harbored other fusion genes involving gene partners previously preventing ligand-receptor interactions,19 such as WIF1.20 WIF1 12 described in PAs, in myoepitheliomas of other anatomical sites consists of an N-terminal secretion signal, five EGF-like domains, a (i.e., EWSR1 and FUS rearrangements),13,14 or in other tumors hydrophilic C-terminus, and a WIF domain, which is required for displaying myoepithelial differentiation (i.e., CRTC1-MAML2 fusion binding to Wnt proteins and for the tumor suppressor properties gene in mucoepidermoid carcinomas or MYB and MYBL1 of WIF1.19 The HMGA2-WIF1 fusion gene identified in AM16 has rearrangements in adenoid cystic carcinoma).9,15 No additional been shown to result in increased HMGA2 expression, presumably

npj Breast Cancer (2019) 6 Published in partnership with the Breast Cancer Research Foundation F. Pareja et al. 3

Fig. 1 Fusion genes involving HMGA2 or PLAG1 and somatic mutations targeting HRAS, PIK3CA, AKT1 and PIK3R1 in breast adenomyoepitheliomas. a Heatmap depicting fusion gene and somatic mutations targeting HRAS Q61, PIK3CA and AKT E17 hotspot loci and PIK3R1 mutations identified in breast adenomyoepitheliomas (AMEs; n = 13). Cases are shown in columns and genes in rows. Hotspot mutations are annotated as per Chang et al.23 b Representative Sanger sequencing electropherograms of HRAS Q61 and PIK3CA hotspot loci in AM16. c Representative hematoxylin and eosin micrographs of an AME harboring an HMGA2-WIF1 fusion gene (AM16), and micrographs depicting p63 and estrogen receptor expression. Scale bars, 500 μm (upper left), 100 μm (upper right) and 50 μm (middle and lower panels). ER estrogen receptor, FISH fluorescence in situ hybridization, SNV single nucleotide variant, WT wild-type due to loss of regulatory sites in its 3′ UTR,21 and decreased WIF1 Taken together, our findings suggest that a subset of AMEs lacking expression.10 The HMGA2-WIF1 chimeric transcript identified in our genetic alterations involving genes of the RAS-MAPK pathway study is predicted to encode a full length HMGA2 protein and an may be underpinned by fusion genes resulting in the activation of N-terminally truncated WIF1 protein harboring a truncated WIF alternative signaling pathways, such as TGF-β and Wnt. domain. Given that the HMGA2 breakpoint maps to its 3’ UTR after One could posit that the AME harboring the HMGA2-WIF1 fusion the stop codon, it is possible that WIF1 may not even be gene described in this study would, in fact, constitute a breast PA. translated, akin to a rearrangement involving the same HMGA2 This case was independently reviewed by five breast pathologists and WIF1 exons previously described in a PA arising in the salivary who concurred in the diagnosis of AME. It should be noted, gland.16 In addition, HMGA2 and WIF1 display opposite transcrip- however, that despite being a bona fide AME, this case focally tional orientations, and this fusion gene may stem from a cryptic displayed increased myxoid stroma, bearing some resemblance to, paracentric inversion (Fig. 2a).16 but not fulfilling the diagnostic criteria for, a breast PA. Another Taken together, the HMGA2-WIF1 fusion identified here might potential explanation for the presence of this fusion gene in AM16 result in increased expression of HMGA2, with ensuing activation is that it would constitute the breast equivalent of the salivary of TGF-β signaling, along with derepression of Wnt signaling. gland epithelial-myoepithelial carcinoma ex-PA.11 The central

Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2019) 6 F. Pareja et al. 4

Fig. 2 HMGA2-WIF1 fusion gene identified in the epithelial and myoepithelial cells of a breast adenomyoepithelioma. a Schematic representation of the HMGA2-WIF1 fusion transcript identified in AM16, including the exons and domains involved. HMGA2 is on the (+) DNA strand and WIF1 on the (−) DNA strand. The breakpoints of the 5’ and 3’ partner genes are represented as black vertical lines. Eight spanning reads were found to cross the genomic breakpoint of the HMGA2-WIF1 chimeric transcript and are depicted aligned to the predicted junction sequence. b Representative hematoxylin and eosin and FISH micrographs of the epithelial and myoepithelial components of AM16 using HMGA2 dual-color break apart probes (red, 5′ HMGA2; green, 3′ HMGA2). aa aminoacid, AcD acidic domain, DBD DNA binding domain, E epithelium, M myoepithelium, SpD spacer domain. Scale bar, 50 μm

pathology review of all slides available from this case failed to METHODS reveal any areas diagnostic of PA. Therefore, our findings suggest Cases and DNA sequencing data that a subset of AMEs share not only morphologic features with This study was approved by the Institutional Review Boards (IRBs) and local PAs, but may also resemble PAs at the genetic level. We cannot research ethics committees of the authors’ institutions. Patient consent rule out, however, that AM16 developed in the context of a PA, was obtained if required by the approved IRB protocols. In this study we which was subsequently obliterated by the outgrowth of the AME. included thirteen breast AMEs, retrieved from the authors’ institutions and 3 The FISH analysis of AM16 revealed the presence of the HMGA2- previously described by Geyer et al. Whole-exome sequencing, MSK- IMPACT and Sanger sequencing data, and immunohistochemical data WIF1 both in the epithelial and myoepithelial cells of the tumor 3 (Fig. 2b). This observation is consistent with the notion that in were retrieved from Geyer et al. ER status was assessed by immunohis- tochemistry according to the current ASCO/CAP guidelines.22 Hotspot AMEs, both the epithelial and myoepithelial components are mutations are annotated as per Chang et al.23 For power calculations, if we neoplastic and clonally related, even though in this AME (AM16), posited that an HRAS-wild type AMEs would be underpinned by a recurrent the epithelial component was ER-positive, whereas the myoe- fusion gene and that this fusion gene would be present in ≥ 70% of cases pithelial component was ER-negative. akin to recurrent fusion genes in other tumor types,9,24–26 sequencing Our study was several limitations, such as the small size of our analysis of five samples would confer 80% power for its detection. cohort and the fact that we included only two HRAS-/PIK3CA-/ AKT1-wild type AMEs, given that no additional material from other RNA-sequencing and the identification of fusion transcripts AMEs was available for transcriptomic or FISH analysis. The limited RNA-sequencing was performed on five HRAS-wild type AMEs according to sample size of our study precludes definitive conclusions standard protocols employed at the Integrated Genomics Operation of regarding the relationship between ER status and the presence Memorial Sloan Kettering Cancer Center (MSKCC).27 In brief, paired-end of the HMGA2-WIF1 fusion gene in AMEs to be drawn. Despite massively parallel RNA-sequencing (2 × 50 bp) was performed on a 28 these limitations, our findings demonstrate that AMEs lacking HiSeq2000 (Illumina), as previously described. Read pairs supporting fi 29 30 mutations affecting HRAS, PIK3CA and AKT1 may harbor the chimeric transcripts were identi ed using deFuse, and INTEGRATE, followed by exclusion of candidate fusion transcripts found in a set of 287 HMGA2-WIF1 fusion gene, previously described in salivary gland 31 28 10,16 normal breast tissues from the TCGA dataset, as previously described. PAs and carcinomas ex-PA, suggesting that a subset of AMEs The Bayesian probability of the remaining candidate fusion genes, may be genetically related to PAs or that AMEs may originate in supported by at least two spanning reads, to constitute drivers was the context of breast PAs. annotated using OncoFuse,32 as previously described.28

npj Breast Cancer (2019) 6 Published in partnership with the Breast Cancer Research Foundation F. Pareja et al. 5 Fluorescence in situ hybridization (FISH) 9. Kim, J. et al. MYBL1 rearrangements and MYB amplification in breast adenoid cystic All cases included in this study (n = 13) were subjected to FISH analysis for carcinomas lacking the MYB-NFIB fusion gene. J. Pathol. 244,143–150 (2018). HMGA2 and PLGA1 using dual-color break-apart probes following validated 10. Queimado, L., Lopes, C. S. & Reis, A. M. WIF1, an inhibitor of the Wnt pathway, is protocols at the MSKCC Molecular Cytogenetics Core, as previously rearranged in salivary gland tumors. Genes Cancer 46,215–225 (2007). described.33 The probe mix consisted of bacterial artificial 11. Antony, J., Gopalan, V., Smith, R. A. & Lam, A. K. Carcinoma ex pleomorphic (BAC) clones mapping to 5′ HMGA2 (RP11-230G5, RP11-662G15; red) and 3′ adenoma: a comprehensive review of clinical, pathological and molecular data. HMGA2 (RP11-937C6, RP11-167E10; green), and BAC clones mapping to 5′ Head. Neck Pathol. 6,1–9 (2012). PLAG1 (RP11-92A9, RP11-111I18; red) and 3′ PLAG1 (RP11-144E19, RP11- 12. Antonescu, C. R. et al. Frequent PLAG1 gene rearrangements in skin and soft 246A9; green). A minimum of 50 interphase nuclei were analyzed for HMGA2 tissue myoepithelioma with ductal differentiation. Genes Chromosomes Cancer or PLAG1 rearrangements. Cases were considered positive for rearrangement 52, 675–682 (2013). if separation of the 5′ (red) and 3′ (green) signals (>2 signal width apart) was 13. Huang, S. C. et al. Novel FUS-KLF17 and EWSR1-KLF17 fusions in myoepithelial identified in >15% tumor cells. FISH analyses were performed with observers tumors. Genes Chromosomes Cancer 54, 267–275 (2015). blinded to the results of the RNA-sequencing analysis. 14. Antonescu, C. R. et al. EWSR1-POU5F1 fusion in soft tissue myoepithelial tumors. A molecular analysis of sixty-six cases, including soft tissue, bone, and visceral lesions, showing common involvement of the EWSR1 gene. Genes Chromosomes DATA AVAILABILITY Cancer 49, 1114–1124 (2010). 15. Tonon, G. et al. t(11;19)(q21; p13) translocation in mucoepidermoid carcinoma RNA-sequencing data that support the findings of this study have been deposited in creates a novel fusion product that disrupts a Notch signaling pathway. Nat. the NCBI Sequence Read Archive (SRA) under the accession code SRP158271. WES Genet. 33, 208–213 (2003). and MSK-IMPACT sequencing data retrieved from Geyer et al3 are available in SRA 16. Persson, F. et al. High-resolution genomic profiling of adenomas and carcinomas under the accession numbers SRP065277 and SRP065302, respectively. All available of the salivary glands reveals amplification, rearrangement, and fusion of HMGA2. data are available from the authors. Genes Chromosomes Cancer 48,69–82 (2009). 17. Fusco, A. & Fedele, M. Roles of HMGA proteins in cancer. Nat. Rev. Cancer 7, 899–910 (2007). ACKNOWLEDGEMENTS 18. Morishita, A. et al. HMGA2 is a driver of tumor metastasis. Cancer Res 73, This study was funded by the Breast Cancer Research Foundation. Research reported 4289–4299 (2013). in this publication was funded in part by a Cancer Center Support Grant of the 19. Malinauskas, T., Aricescu, A. R., Lu, W., Siebold, C. & Jones, E. Y. Modular National Institutes of Health/National Cancer Institute (grant No P30CA008748). The mechanism of Wnt signaling inhibition by Wnt inhibitory factor 1. Nat. Struct. Mol. content is solely the responsibility of the authors and does not necessarily represent Biol. 18, 886–893 (2011). the official views of the National Institutes of Health. 20. Hsieh, J. C. et al. A new secreted protein that binds to Wnt proteins and inhibits their activities. Nature 398, 431–436 (1999). 21. Mayr, C., Hemann, M. T. & Bartel, D. P. Disrupting the pairing between let-7 and AUTHORS’ CONTRIBUTIONS Hmga2 enhances oncogenic transformation. Science 315, 1576–1579 (2007). B.W. and J.S.R.-F. conceived the study. H.Y.W., Z.V., J.P., B.P.R., I.O.E., E.B. and E.A.R. 22. Hammond, M. E. et al. American Society of Clinical Oncology/College Of American provided tissue samples. F.P., F.C.G, M.E., A.P.M.S., E.A.R., I.O.E. and J.S.R.-F. performed Pathologists guideline recommendations for immunohistochemical testing of estro- – the pathology review. Data acquisition, interpretation and analysis were performed gen and progesterone receptors in breast cancer. J. Clin. Oncol. 28, 2784 2795 (2010). by F.P., F.C.G., D.N.B., A.L., A.P.M.S., R.G.-M., A.d.C.P., P.S. and A.A.J. The manuscript was 23. Chang, M. T. et al. Identifying recurrent mutations in cancer reveals widespread initially drafted by F.P. and J.S.R.-F., and all authors edited and approved the final lineage diversity and mutational specificity. Nat. Biotechnol. 34, 155–163 (2016). draft of the manuscript. F.P. and F.C.G. contributed equally to this work. 24. Martelotto, L. G. et al. Genomic landscape of adenoid cystic carcinoma of the breast. J. Pathol. 237, 179–189 (2015). 25. Wetterskog, D. et al. Adenoid cystic carcinomas constitute a genomically distinct ADDITIONAL INFORMATION subgroup of triple-negative and basal-like breast cancers. J. Pathol. 226,84–96 (2012). 26. Ashworth, A., Lord, C. J. & Reis-Filho, J. S. Genetic interactions in cancer pro- Supplementary information accompanies the paper on the npj Breast Cancer gression and treatment. Cell 145,30–38 (2011). website (https://doi.org/10.1038/s41523-018-0101-7). 27. Weinreb, I. et al. Hotspot activating PRKD1 somatic mutations in polymorphous low- grade adenocarcinomas of the salivary glands. Nat. Genet. 46,1166–1169 (2014). Competing interests: J.S.R.-F. reports personal/consultancy fees from VolitionRx, 28. Piscuoglio, S. et al. Integrative genomic and transcriptomic characterization of Page.AI, Goldman Sachs, Grail, Ventana Medical Systems and Genentech, outside the papillary carcinomas of the breast. Mol. Oncol. 8, 1588–1602 (2014). submitted work. The remaining authors declare no competing interests. 29. McPherson, A. et al. deFuse: an algorithm for gene fusion discovery in tumor RNA-Seq data. PLoS Comput. Biol. 7, e1001138 (2011). Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims 30. Zhang, J. et al. INTEGRATE: gene fusion discovery using whole genome and in published maps and institutional affiliations. transcriptome data. Genome Res. 26, 108–118 (2015). 31. Cancer Genome Atlas N. Comprehensive molecular portraits of human breast tumours. Nature 490,61–70 (2012). REFERENCES 32. Shugay, M., Ortiz de Mendibil, I., Vizmanos, J. L. & Novo, F. J. Oncofuse: a com- 1. Yoon, J. Y. & Chitale, D. Adenomyoepithelioma of the breast: a brief diagnostic putational framework for the prediction of the oncogenic potential of gene review. Arch. Pathol. Lab. Med. 137, 725–729 (2013). fusions. Bioinformatics 29, 2539–2546 (2013). 2. Lakhani S. R., Ellis I. O., Schnitt S. J., Tan P. H., van de Vijver M. J. WHO Classification 33. Piscuoglio, S. et al. Uterine adenosarcomas are mesenchymal neoplasms. J. of Breast Tumors. (International Agency for Research Cancer (IARC), Lyon, 2012). Pathol. 238, 381–388 (2016). 3. Geyer, F. C. et al. Recurrent hotspot mutations in HRAS Q61 and PI3K-AKT pathway genes as drivers of breast adenomyoepitheliomas. Nat. Commun. 9, 1816 (2018). Open Access This article is licensed under a Creative Commons 4. McLaren, B. K., Smith, J., Schuyler, P. A., Dupont, W. D. & Page, D. L. Adeno- Attribution 4.0 International License, which permits use, sharing, myoepithelioma: clinical, histologic, and immunohistologic evaluation of a series adaptation, distribution and reproduction in any medium or format, as long as you give of related lesions. Am. J. Surg. Pathol. 29, 1294–1299 (2005). appropriate credit to the original author(s) and the source, provide a link to the Creative 5. Foschini, M. P. et al. The morphological spectrum of salivary gland type tumours Commons license, and indicate if changes were made. The images or other third party of the breast. Pathology 49, 215–227 (2017). material in this article are included in the article’s Creative Commons license, unless 6. Stenman, G. Fusion oncogenes in salivary gland tumors: molecular and clinical indicated otherwise in a credit line to the material. If material is not included in the ’ consequences. Head. Neck Pathol. 7(Suppl 1), S12–S19 (2013). article s Creative Commons license and your intended use is not permitted by statutory 7. Bahrami, A., Dalton, J. D., Krane, J. F. & Fletcher, C. D. A subset of cutaneous and regulation or exceeds the permitted use, you will need to obtain permission directly soft tissue mixed tumors are genetically linked to their salivary gland counterpart. from the copyright holder. To view a copy of this license, visit http://creativecommons. Genes Chromosomes Cancer 51, 140–148 (2012). org/licenses/by/4.0/. 8. Pareja, F. et al. Pleomorphic adenoma and mucoepidermoid carcinoma of the breast are underpinned by fusion genes. Mod. Pathol. 31,98–98 (2018). © The Author(s) 2019

Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2019) 6